1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
//! SPIR-V Instrinics
//!
//! This module is intended as a low level abstraction over SPIR-V instructions.
//! These functions will typically map to a single instruction, and will perform
//! no additional safety checks beyond type-checking.
#[cfg(target_arch = "spirv")]
use crate::integer::Integer;
use crate::{
    integer::{SignedInteger, UnsignedInteger},
    scalar::Scalar,
    vector::Vector,
};
#[cfg(target_arch = "spirv")]
use core::arch::asm;

mod atomics;
mod barrier;
mod demote_to_helper_invocation_ext;
mod derivative;
mod primitive;
mod ray_tracing;

pub use atomics::*;
pub use barrier::*;
pub use demote_to_helper_invocation_ext::*;
pub use derivative::*;
pub use primitive::*;
pub use ray_tracing::*;

/// Result is true if any component of `vector` is true, otherwise result is
/// false.
#[spirv_std_macros::gpu_only]
#[doc(alias = "OpAny")]
#[inline]
pub fn any<V: Vector<bool, N>, const N: usize>(vector: V) -> bool {
    let mut result = false;

    unsafe {
        asm! {
            "%bool = OpTypeBool",
            "%vector = OpLoad _ {vector}",
            "%result = OpAny %bool %vector",
            "OpStore {result} %result",
            vector = in(reg) &vector,
            result = in(reg) &mut result
        }
    }

    result
}

/// Result is true if all components of `vector` is true, otherwise result is
/// false.
#[spirv_std_macros::gpu_only]
#[doc(alias = "OpAll")]
#[inline]
pub fn all<V: Vector<bool, N>, const N: usize>(vector: V) -> bool {
    let mut result = false;

    unsafe {
        asm! {
            "%bool = OpTypeBool",
            "%vector = OpLoad _ {vector}",
            "%result = OpAll %bool %vector",
            "OpStore {result} %result",
            vector = in(reg) &vector,
            result = in(reg) &mut result
        }
    }

    result
}

/// Extract a single, dynamically selected, component of a vector.
///
/// # Safety
/// Behavior is undefined if `index`’s value is greater than or equal to the
/// number of components in `vector`.
#[spirv_std_macros::gpu_only]
#[doc(alias = "OpVectorExtractDynamic")]
#[inline]
pub unsafe fn vector_extract_dynamic<T: Scalar, const N: usize>(
    vector: impl Vector<T, N>,
    index: usize,
) -> T {
    let mut result = T::default();

    asm! {
        "%vector = OpLoad _ {vector}",
        "%element = OpVectorExtractDynamic _ %vector {index}",
        "OpStore {element} %element",
        vector = in(reg) &vector,
        index = in(reg) index,
        element = in(reg) &mut result
    }

    result
}

/// Make a copy of a vector, with a single, variably selected,
/// component modified.
///
/// # Safety
/// Behavior is undefined if `index`’s value is greater than or equal to the
/// number of components in `vector`.
#[spirv_std_macros::gpu_only]
#[doc(alias = "OpVectorInsertDynamic")]
#[inline]
pub unsafe fn vector_insert_dynamic<T: Scalar, V: Vector<T, N>, const N: usize>(
    vector: V,
    index: usize,
    element: T,
) -> V {
    let mut result = V::default();

    asm! {
        "%vector = OpLoad _ {vector}",
        "%element = OpLoad _ {element}",
        "%new_vector = OpVectorInsertDynamic _ %vector %element {index}",
        "OpStore {result} %new_vector",
        vector = in(reg) &vector,
        index = in(reg) index,
        element = in(reg) &element,
        result = in(reg) &mut result,
    }

    result
}

/// Fragment-shader discard. Equivalvent to `discard()` from GLSL
///
/// Ceases all further processing in any invocation that executes it: Only
/// instructions these invocations executed before [kill] have observable side
/// effects.
#[spirv_std_macros::gpu_only]
#[doc(alias = "OpKill", alias = "discard")]
#[allow(clippy::empty_loop)]
pub fn kill() -> ! {
    unsafe { asm!("OpKill", options(noreturn)) }
}

/// Read from the shader clock with either the `Subgroup` or `Device` scope.
///
/// See:
/// <https://htmlpreview.github.io/?https://github.com/KhronosGroup/SPIRV-Registry/blob/master/extensions/KHR/SPV_KHR_shader_clock.html>
#[cfg(all(
    target_feature = "Int64",
    target_feature = "ShaderClockKHR",
    target_feature = "ext:SPV_KHR_shader_clock"
))]
#[spirv_std_macros::gpu_only]
#[doc(alias = "OpReadClockKHR")]
pub unsafe fn read_clock_khr<const SCOPE: u32>() -> u64 {
    let mut result: u64;

    asm! {
        "%uint = OpTypeInt 32 0",
        "%scope = OpConstant %uint {scope}",
        "{result} = OpReadClockKHR typeof*{result} %scope",
        result = out(reg) result,
        scope = const SCOPE,
    };

    result
}

/// Like `read_clock_khr` but returns a vector to avoid requiring the `Int64`
/// capability. It returns a 'vector of two-components of 32-bit unsigned
/// integer type with the first component containing the 32 least significant
/// bits and the second component containing the 32 most significant bits.'
#[cfg(all(
    target_feature = "ShaderClockKHR",
    target_feature = "ext:SPV_KHR_shader_clock"
))]
#[spirv_std_macros::gpu_only]
#[doc(alias = "OpReadClockKHR")]
pub unsafe fn read_clock_uvec2_khr<V: Vector<u32, 2>, const SCOPE: u32>() -> V {
    let mut result = V::default();

    asm! {
        "%uint = OpTypeInt 32 0",
        "%scope = OpConstant %uint {scope}",
        "%result = OpReadClockKHR typeof*{result} %scope",
        "OpStore {result} %result",
        result = in(reg) &mut result,
        scope = const SCOPE,
    };

    result
}

#[cfg(target_arch = "spirv")]
unsafe fn call_glsl_op_with_ints<T: Integer, const OP: u32>(a: T, b: T) -> T {
    let mut result = T::default();
    asm!(
        "%glsl = OpExtInstImport \"GLSL.std.450\"",
        "%a = OpLoad _ {a}",
        "%b = OpLoad _ {b}",
        "%result = OpExtInst typeof*{result} %glsl {op} %a %b",
        "OpStore {result} %result",
        a = in(reg) &a,
        b = in(reg) &b,
        result = in(reg) &mut result,
        op = const OP
    );
    result
}

/// Compute the minimum of two unsigned integers via a GLSL extended instruction.
#[spirv_std_macros::gpu_only]
pub fn unsigned_min<T: UnsignedInteger>(a: T, b: T) -> T {
    unsafe { call_glsl_op_with_ints::<_, 38>(a, b) }
}

/// Compute the maximum of two unsigned integers via a GLSL extended instruction.
#[spirv_std_macros::gpu_only]
pub fn unsigned_max<T: UnsignedInteger>(a: T, b: T) -> T {
    unsafe { call_glsl_op_with_ints::<_, 41>(a, b) }
}

/// Compute the minimum of two signed integers via a GLSL extended instruction.
#[spirv_std_macros::gpu_only]
pub fn signed_min<T: SignedInteger>(a: T, b: T) -> T {
    unsafe { call_glsl_op_with_ints::<_, 39>(a, b) }
}

/// Compute the maximum of two signed integers via a GLSL extended instruction.
#[spirv_std_macros::gpu_only]
pub fn signed_max<T: SignedInteger>(a: T, b: T) -> T {
    unsafe { call_glsl_op_with_ints::<_, 42>(a, b) }
}

/// Index into an array without bounds checking.
///
/// The main purpose of this trait is to work around the fact that the regular `get_unchecked*`
/// methods do not work in in SPIR-V.
pub trait IndexUnchecked<T> {
    /// Returns a reference to the element at `index`. The equivalent of `get_unchecked`.
    ///
    /// # Safety
    /// Behavior is undefined if the `index` value is greater than or equal to the length of the array.
    unsafe fn index_unchecked(&self, index: usize) -> &T;
    /// Returns a mutable reference to the element at `index`. The equivalent of `get_unchecked_mut`.
    ///
    /// # Safety
    /// Behavior is undefined if the `index` value is greater than or equal to the length of the array.
    unsafe fn index_unchecked_mut(&mut self, index: usize) -> &mut T;
}

impl<T> IndexUnchecked<T> for [T] {
    #[cfg(target_arch = "spirv")]
    unsafe fn index_unchecked(&self, index: usize) -> &T {
        // FIXME(eddyb) `let mut result = T::default()` uses (for `asm!`), with this.
        let mut result_slot = core::mem::MaybeUninit::uninit();
        asm! {
            "%slice_ptr = OpLoad _ {slice_ptr_ptr}",
            "%data_ptr = OpCompositeExtract _ %slice_ptr 0",
            "%result = OpAccessChain _ %data_ptr {index}",
            "OpStore {result_slot} %result",
            slice_ptr_ptr = in(reg) &self,
            index = in(reg) index,
            result_slot = in(reg) result_slot.as_mut_ptr(),
        }
        result_slot.assume_init()
    }

    #[cfg(not(target_arch = "spirv"))]
    unsafe fn index_unchecked(&self, index: usize) -> &T {
        self.get_unchecked(index)
    }

    #[cfg(target_arch = "spirv")]
    unsafe fn index_unchecked_mut(&mut self, index: usize) -> &mut T {
        // FIXME(eddyb) `let mut result = T::default()` uses (for `asm!`), with this.
        let mut result_slot = core::mem::MaybeUninit::uninit();
        asm! {
            "%slice_ptr = OpLoad _ {slice_ptr_ptr}",
            "%data_ptr = OpCompositeExtract _ %slice_ptr 0",
            "%result = OpAccessChain _ %data_ptr {index}",
            "OpStore {result_slot} %result",
            slice_ptr_ptr = in(reg) &self,
            index = in(reg) index,
            result_slot = in(reg) result_slot.as_mut_ptr(),
        }
        result_slot.assume_init()
    }

    #[cfg(not(target_arch = "spirv"))]
    unsafe fn index_unchecked_mut(&mut self, index: usize) -> &mut T {
        self.get_unchecked_mut(index)
    }
}

impl<T, const N: usize> IndexUnchecked<T> for [T; N] {
    #[cfg(target_arch = "spirv")]
    unsafe fn index_unchecked(&self, index: usize) -> &T {
        // FIXME(eddyb) `let mut result = T::default()` uses (for `asm!`), with this.
        let mut result_slot = core::mem::MaybeUninit::uninit();
        asm! {
            "%result = OpAccessChain _ {array_ptr} {index}",
            "OpStore {result_slot} %result",
            array_ptr = in(reg) self,
            index = in(reg) index,
            result_slot = in(reg) result_slot.as_mut_ptr(),
        }
        result_slot.assume_init()
    }

    #[cfg(not(target_arch = "spirv"))]
    unsafe fn index_unchecked(&self, index: usize) -> &T {
        self.get_unchecked(index)
    }

    #[cfg(target_arch = "spirv")]
    unsafe fn index_unchecked_mut(&mut self, index: usize) -> &mut T {
        // FIXME(eddyb) `let mut result = T::default()` uses (for `asm!`), with this.
        let mut result_slot = core::mem::MaybeUninit::uninit();
        asm! {
            "%result = OpAccessChain _ {array_ptr} {index}",
            "OpStore {result_slot} %result",
            array_ptr = in(reg) self,
            index = in(reg) index,
            result_slot = in(reg) result_slot.as_mut_ptr(),
        }
        result_slot.assume_init()
    }

    #[cfg(not(target_arch = "spirv"))]
    unsafe fn index_unchecked_mut(&mut self, index: usize) -> &mut T {
        self.get_unchecked_mut(index)
    }
}